Task 3: Irpinia Fault SystemWP3.1 Seismic noise analysis and

Green Functions

Project – DPC S5High-resolution multi-disciplinary

monitoring of active fault test-site areas in Italy

Vassallo Maurizio, Gaetano Festa, Antonella Bobbio, Piero Brondi

24 March 2010 –INGV Rome

Broad-band ISNet stations and Ambient seismic Noise

RSF3

RDM3

PGN3

COL3

TEO3

Ambient seismic noise acquired for 18 months at 5 stations of ISNet equipped with broad-band velocimeters (Trillium 40)

ISNet dense seismic network:• A field laboratory to study the seismic source at small scales• An advanced infrastructure to test early warning procedures

Broad-band stations:RSF3 Rocca S. Felice (AV)TEO3 Teora (AV)RDM3 Ruvo del Monte(PZ)COL3 Colliano (SA)PGN3 Pignola (PZ)

Green functions from ambient seismic noise

and ФA and ФB are the seismic fields received by two sensors in A and B

When the propagation medium satisfies the equipartition principle for a complete diffuse wavefield we can compute the Green function GAB(t) between two recording points A and B by the cross-correlation functions of the respectively seismic fields:

GAB(t) can be reitrieved by the causal part (t > 0) and anticausal part (t<0) of the cross correlation function CAB(t)

Lobkis and Weaver (2001)

Where

Data collecting and processing

Phase 2: cross-correlation and stack

Continuous seismic data acquired for 18 months at 5 broad band stations of ISNet

Remove mean; remove trend; band pass filter and cut to

length 6 hours

time domain normalization (1-bit normalization)

Spectral Normalization

(spectral whitening)

Compute cross-correlation

Compute the stack of correlations

Phase 1: single station data preparation

Cross-correlations and stacks

Common low frequency signal

Higher stability of signal during

the spring and summer

• symmetric stack• energetic signal between -30 e 30 s

• asymmetric stack• energetic signal between -70 e 70 s

Distance between stations 27 km |One year of data (2009) | Distance between stations 40 km

small variations in the processed stacks processing

not effective

Processing to increase signal quality•Spectral whitening •Recoursive Butterworth filter between 0.1 Hz and 1 Hz•Stack of traces with high S/N

Whitening Recoursive filter

Dispersion analysis [0.1Hz-1Hz]

frequency (Hz)Gro

up v

eloc

ity (m

/s)

Period (s)Gro

up v

eloc

ity (k

m/s

)

PGN-RSF

PGN-TEO

Gro

up v

eloc

ity (k

m/s

)

Period (s)

Surface wave analysis of stacks•Velocity analysis for identification of phases •Picking for dispersion curves reconstruction

Velocity models•Comparison between picked dispersion curves and dispersion curves computed using the 1D velocity models of the area

Dispersion analysis [0.1Hz-1Hz]

Dispersion computed only for far stations pairs ,for near stations pairs low S/N in stack traces problems for velocity analysis

Dispersion analysis [10s-50s]

All stacks filtered in 0.05Hz-0.08 Hz

PGN-RSF0.01 Hz0.02 Hz0.03 Hz0.04 Hz0.05 Hz0.06 Hz0.07 Hz0.08 Hz0.09 Hz0.10 Hz

Time-dispersion analysis

Surface wave propagation

Frequency

Distance Period (s)

Gro

up v

eloc

ity (k

m/s

)

RDM-RSF

Period (s)

Gro

up v

eloc

ity (

km/s

)

RDM-PGN

Velocity analysis

Preliminary

results

Dispersion analysis [10s-50s]

Preliminary

results

Clear surface waves identified on stack traces for all broad band stations

Conclusions

• Data collections and processing 100% completed • Computed cross-Correlations and Green functions for

all pairs of broad band stations 100% completed• Velocity and dispersion analysis 80-90% completed • Inversion dispersion curves for reconstruction of

velocity models 30-40% completed

WP 3.1 activities24

months18

months12

months6

months

Data collection and processing

Green functions

Dispersion analysis

Velocity models

Time Schedule